Tomographic Reconstruction¶

Overview¶

This interface aims at integrating and simplifying the following tasks related to tomographic reconstruction and imaging with neutrons. While much of its functionality is being developed in a generic way, it is presently being tested and trialed for the IMAT instrument at ISIS.

An important feature of this interface is the capability to submit jobs to a remote compute resource (a compute cluster for example). Currently remote jobs are run on the SCARF cluster, administered by the Scientific Computing Department of STFC. You can also use this cluster via remote login and through its web portal. This resource is available for ISIS users.

Interface at a glance¶

By default the interface shows the Run tab, where you can visualize images, submit reconstruction jobs, see and control the status of the jobs submitted recently.

In the setup tab you can set the details of the remote and/or local compute resources. Importantly, here is where you can set you username and password to log into the remote compute resource. To be able to run jobs remotely you first need to log into the remote compute resource. Once you log in, an automatic mechanism will periodically query the status of jobs (for example every minute). You can also update it at any time by clicking on the refresh button.

In this tab you also have to set the folders/directories where the input data for reconstruction jobs is found. This information is required every time you start analyzing a new dataset. The required fields are:

Samples directory

Directory containing the sample images

Open beam directory

Where to find the open beam (flat/white) image(s)

Dark field directory

Where to find the dark image(s)

NB: This interface is under heavy development. Several practical details lack polishing and/or are missing. This implies that there may be usability issues at times and some steps may not be as intuitive or simple as they could. Please, do not hesitate to provide suggestions and feedback.

The next sections provide further details that might be needed to fully understand the process of generating tomographic reconstructions with this interface.

Tools¶

At the moment two reconstruction tools are being set up and trialed on SCARF and some ISIS machines:

• TomoPy
• Astra Toolbox found from here.

References for the Astra Toolbox:

• W. van Aarle, W J. Palenstijn, J. De Beenhouwer, T. Altantzis, S. Bals, K. J. Batenburg, and J. Sijbers, “The ASTRA Toolbox: a platform for advanced algorithm development in electron tomography”, Ultramicroscopy, Vol. 147, p. 35–47, (2015)
• W J. Palenstijn, K J. Batenburg, and J. Sijbers, “Performance improvements for iterative electron tomography reconstruction using graphics processing units (GPUs)”, Journal of structural biology, vol. 176, issue 2, pp. 250-253, 2011

References for TomoPy:

• Gursoy D, De Carlo F, Xiao X, Jacobsen C. (2014). TomoPy: a framework for the analysis of synchrotron tomographic data. J. Synchrotron Rad. 21. 1188-1193 doi:10.1107/S1600577514013939

In the near future it is expected that support will be added for Savu: Tomography Reconstruction Pipeline, developed at the Diamond Light Source.

References for Savu:

• Atwood R C, Bodey A J, Price S W T, Basham M and Drakopoulos M 2015 A high-throughput system for high-quality tomographic reconstruction of large datasets at diamond light source Philosophical Transactions A 373 20140398

Data formats¶

In principle, users do not need to deal with specificities of different file formats. That is the aim of this interface, but as it is currently being developed, and for reference a brief list of relevant file and data formats is given here:

• FITS: Flexible Image Transport System format used to store images in files. You can see the details on how FITS images can be loaded into Mantid in the documentation of the algorithm LoadFITS http://docs.mantidproject.org/nightly/algorithms/LoadFITS-v1.html.
• TIFF: Tagged Image File Format images used as FITS for image or slice files. This format is presently not supported in the Mantid data analysis framework but it is used in the tomographic reconstruction interface.
• Diamond Light Source (DLS) NXTomo: a specific NeXus format used by some of the tools that this interface supports or will support. See next sections for details.

These formats are used in different processing steps and parts of this interface. For example, you can visualize FITS and TIFF images in the Run tab and also in the ROI, etc. tab. As another example, the reconstruction tools typically need as inputs at least a stack of images which can be in different formats, including a set of FITS or TIFF files, or a single DLS NXTomo file. Other third party tools use files in these formats as inputs, outputs or both.

Data locations¶

This is dependent on the facility and instrument.

TODO: this is work in progress. At ISIS In principle data will be replicated in the ISIS archive, the SCARF imat disk space, and possibly an analysis machine located in R3.

The tab Visualization has simple push buttons to browse the files available from the local and remote locations, as well as any other directory or folder selected by the user.

Running jobs remotely¶

To be able to run jobs on a remote compute resource (cluster, supercomputer, etc.)

• Log into the resource
• Select it and setup one reconstruction tool
• Use the reconstruct button in the Run tab of the interface

You can monitor the status of the jobs currently running (and recently run) on remote compute resources in the same tab.

Setting common parameters for the reconstruction jobs¶

Before any reconstruction job is started several pre-/post-processing options would normally need to be fine tuned for the sample data to be processed correctly. The region of interest and the “air” region (or region for normalization) can be set visually in a specific tab. All other pre- and post-processing settings are defined in a separate tab.

Regions¶

Several parameters can be set in the ROI etc. section or tab. These parameters will be used for all the reconstruction jobs, regardless of the tool and/or reconstruction method used.

• Region of interest (ROI) for the analysis
• Area for normalization (open beam, not blocked by sample)
• Center of rotation, for tomographic reconstruction

At any stage during the process of selecting the regions it is also possible to see how the selections fit different images by sliding through the images of the stack (using the slider or scroll bar).

The center of rotation can be selected interactively by clicking on the select button and then clicking on an image pixel. To select the regions of interest or the area of normalization, just click on the respective “select” button and then click and drag with the mouse to select a rectangle. The precise coordinates of the center and regions can be set via the boxes of the right panel as well.

Once you have selected or set one of the regions, or the center, they can be selected again by pushing the respective “Select” buttons and/or editing their coordinates manually.

The default values, set in principle when a new stack of images is loaded, is as follows. The region of intererest is set to cover all the images. The regions of normalization is not set (empty), and the center of rotation is set to the center of the image. The option to find the center of rotation automatically is disabled at present.

If when selection a region the mouse is moved outside of the images, it is possible to continue the selection of the region (second corner) by clicking again inside the image. Alternatively, any selection can be reset at any point by using the “reset” buttons.

Pre-/post-processing¶

The Filters tab can be used to set up the pre- and post-processing steps. These are applied regardless of the particular tomographic reconstruction tool and algorithm used when running reconstruction jobs. Pre-processing filters are applied on the raw input images before the reconstruction algorithm is run. Post-processing steps are applied on the reconstructed volume produced by the algorithm.

The tab also shows options to define what outputs should be produced in addition to the reconstructed volume.

The settings are remembered between sessions. It is possible to reset all the settings to their original defaults by clicking on the reset button.

Results from reconstruction jobs¶

The results are written into the output paths selected in the interface (in the setup section or tab). For every reconstructed volume a sequence of images (slices along the vertical axis) are written. In addition, two complementary outputs are generated in the same location:

• A readme file with detailed information on the reconstruction and settings (0.README_reconstruction.txt), including paths, algorithms, filters and parameters used.
• A compressed package file that contains the scripts and subpackages used for the reconstruction job, for reproducibility and to make it easier to track down potential issues. This file is written as 0.reconstruction_scripts.zip

Running jobs locally¶

This capability is being developed at the moment, and it requires additional setup steps on the local analysis machine. Basic functionality is supported only on the IMAT data analysis machine.

Visualization¶

The Visualization tab can be used to browse the local and remote locations where results are stored. It is also possible to open these results in third party visualization applications. NB: ParaView is currently supported and additional tools are being integrated.

Energy bands¶

Conversion between formats¶

Example¶

TODO: there should be an example using a small data set.

TomoPy¶

TODO: how to use it. Hints.

Astra Toolbox¶

TODO: how to use it. Hints.

Astra Toolbox¶

TODO: how to use it. Hints.

Savu¶

TODO: how to use it. Hints.

Savu uses a specific file format developed by the Diamond Light Source, the DLS NXTomo. A few examples can be found from the savu repository on GitHub.